Object Databases Overview

Object Databases

• Object databases store complex objects, which consist of an identifier and a state.
• Object identifiers are immutable, while the state can be modified.
• Literals have a fixed value but lack an object identifier.
• Transient objects are temporary and exist only during program execution.
• Persistent objects are permanently stored in the database and survive program termination.
• Naming mechanism allows objects to be found by their unique names.
• Reachability enables objects to be located through references from other persistent objects.
• Inheritance allows the definition of new object types (subtypes) based on existing types (supertypes), creating a hierarchy of types or classes.
• Extents are named persistent objects that hold collections of all persistent objects for a specific class.
• Every object in a subtype's extent must also be a member of its supertype's extent.
• Objects are characterized by five aspects: identifier, name, lifetime, structure, and creation.
• Behavior inheritance, also known as "IS-A" or interface inheritance, allows subtypes to inherit the behavior of their supertypes.

File Organization

• Fixed-length records are all the same size within a file.
• Variable-length records have different sizes, and can be stored using spanned or unspanned organization.
• Spanned organization allows records to span multiple blocks when they exceed a single block size.
• Unspanned organization restricts records to a single block boundary.
• Dense indexes contain an index entry for each search key value in the data file.
• Sparse indexes contain index entries for only a subset of search key values.
• Secondary indexes provide an alternative access mechanism for data files.
• Index tuning, also known as index optimization, involves reviewing and adjusting database indexes to improve query performance.

Indexing & Performance

• Slow queries can occur due to the absence of suitable indexes.
• Unused indexes consume resources unnecessarily.
• Frequent updates to indexed data can decrease database performance.
• Index tuning needs dynamic evaluation due to the varying nature of database usage over time.
• Creating or dropping indexes can improve performance but can temporarily disrupt table updates.

Concurrency Control

• Binary locks offer two states: locked and unlocked (1 and 0).
• Read locks (share locks) allow multiple transactions to read an item concurrently.
• Write locks (exclusive locks) grant exclusive read/write access to a single transaction.
• Lock upgrade allows a transaction to transition from a read lock to a write lock if it's the only reader.
• Lock downgrade allows a transaction to transition from a write lock to a read lock.
• A serial schedule ensures that transactions execute one after another in a strict order.
• Deadlock occurs when two or more transactions wait for each other's locked resources.
• Starvation occurs when a transaction remains blocked for an extended period while other transactions proceed.
• When a deadlock is detected, a "victim" transaction is chosen to be terminated.
• Choosing long-running or high-update transactions as victims should be avoided to prevent significant data loss.